Computation and Interpretation in Literary Studies

Abstract

The article suggests that the best examples of textual work in the computational humanities are best understood as motivated by aesthetic concerns with the constraints placed on literature by computation’s cultural hegemony. To draw these concerns out, I adopt a middle-distant depth of field, examining the strange epistemology and unexpected aesthetic dimension of numerical culture’s encounters with literature. The middle-distant forms of reading I examine register problematically as literary scholarship not because they lack rigor or evidence but because their unacknowledged object of study is the infrastructure of academic knowledge production. Work in the computational humanities is approaching a point at which the scale of analyzed data and data analysis washes out readings, the algorithms are achieving opaque complexity, and the analytical systems are producing purposive outputs. These problems cannot be addressed without attending to the aesthetics of data-driven cultural encounters, specifically the questions of how we produce readings/viewings and how they change our perceptions and characterize the interesting, critical theorization on method and meaning that make the best work in the computational humanities legitimately humanistic. I contribute a working example: a recommendation system for passages within the Shakespearean dramatic corpus, built using a large bibliographical dataset from JSTOR, a counting/ranking algorithm used at large scale. The system returns passages as intertexts for the passage a reader has selected. I explain how and why this system provides meaningful intertextual connections within the Shakespearean dramatic corpus by tracing the legible structural effects of disciplinary knowledge formation on the shape of this dataset. I close by suggesting how the computational and more traditional methods in the humanities might begin to stop debating past one another.